Bayesian Linear Time Software Agents for Supporting Faster than Real-time Course Of Action Analysis

We propose research to extend Management Sciences’ existing Cognitive Bayesian (CBNX) architecture to create a family of Situation Aware Reflective Cognitive Agents (SARCA) supporting real time information management and cognitive information fusion. This framework will enable development of high performance Modeling and Simulation (M&S) applications fusing data and information from operational data sources running on a High Performance Computing (HPC) platform. The SARCA shall operate as a distributed agent team using probabilistic inference and learning to anticipate, plan, schedule and connect data sources in response to dynamic M&S requirements. The SARCA will be situation aware, using a series of context frames to provide a framework that addresses real-life Command, Control, Communications, Computers, Intelligence, Surveillance, and Reconnaissance (C4ISR) effects challenges in support of faster than real-time Course Of Action Analysis (COAA). Because each SARCA is both self- and situation-aware, it is capable of dynamically adapting its model composition based upon its situation and task requirements. SARCAs are capable of incorporating dynamic extensions to their local ontology, and can exploit this capability in communication and collaboration with other SARCAs and with supervisory agents in solving bandwidth and reliability issues.